Canon 5d mark iii improved- olpf removed for higher resolution

Emanuel said:
On the spot ;-) Let's stop to be naive. They want to protect their high-end to our own expenses.

http://www.dvxuser.com/V6/showthrea...ppling&p=1986113437&viewfull=1#post1986113437
http://www.dvxuser.com/V6/showthrea...ppling&p=1986114792&viewfull=1#post1986114792
Click to expand...

Yep!

Emanuel said:
from Vimeo:

Q: Secondary OLPF?
A: It has two of them. This is just the outer one that doubles as the self cleaning plate. An IR cut filter and 1st OLPF are fixed on the sensor assembly. Not removed here.

https://vimeo.com/39541733
Click to expand...

Added the second one to the self cleaning plate...
 
PappasArts said:
I got a lengthy email from someone about all this. I wasn't aware the DIGIC 5 debayered. Cameras like RED it's done in SW, however I guess it's quite possible that DIGIC 5 does it for video. All the DIGIC chips whether in the C300 or 5DMIII share technical similarities I'm sure. Any opinions on this?

Here is an excerpt:
"Are you aware that the Digic5 debayer's, and this is the reason that there is no increase of moire with the removal of the OLPF from the 5DMIII. Aliasing is not there either because the Digic5 now properly down scales in real time. So you need to ask yourself why the strong additional OLPF then. The answer should be clear."
Click to expand...

There's still aliasing with the OLPF in place. I think it was an engineering & management decision, not protectionism.

H.264 for the 5D3 uses YUV (or more correctly YCrCb): this requires de-Bayering (same thing for the 5D2). The DIGIC5+ is more powerful, and has enough power to perform a filtered downscale, where the prior DIGIC processor required line skipping before processing.

The OLPF(s) need to cut off frequencies higher than the Bayer sensor array. What is strange is that a full res still downsampled to 1920x1080 resolves 1000+ lines. I'd like to see what happens by just disconnecting the 2nd OLPF connector (and leaving the OLPF in place, in the event the firmware behaves differently when the connector is removed). Then again, this increased sharpness may have something to do with the different Bayer sampling and binning for video.

The binning for video acts like a low pass filter (2x2 (green), 3x3 (r&b), my guess). Thus while aliasing my be present on the photosites with the 2nd OLPF removed, after averaging (low-pass filtering) the aliasing is reduced or removed. After writing this out, I suspect that's why it works without a ton of aliasing and moire for video (so far :)). If aliasing and moire don't increase dramatically (up to about the same level as a C300, for example), then it may be worth the $450 and warranty voiding to have the OLPF replaced with a piece of optical glass by maxmax.com. That said, only if I can't get close with post-processing (then maybe after warranty expires might be worth it).
 
Last edited:
jcs said:
I'd like to see what happens by just disconnecting the 2nd OLPF connector (and leaving the OLPF in place, in the event the firmware behaves differently when the connector is removed).
Click to expand...

What connector? An OLPF is a piece of glass, no electronics involved. That's the the "O" is for - Optical Low Pass Filter.
 
Last edited:
Maybe the causation is in the other direction, John. Is it possible that the OLPF is an electro-optic, much like LCD? Why else would it have a connector? Can the corner frequency of the low-pass roll-off be altered with electric current or charge?
 
nothing- see the tear down pics- there's a connector.
bernard- connector is for dust vibration cleaning system.

After thinking about this further, I think the reason we're not seeing more aliasing is due to the low-pass filtering effect of pixel binning the Bayer array.
 
conector is for dust filter
second olpf is part of that ...
only mathematically perfect olpf will cut only moire and alliasing ...
but in real life that is not a case ....
how strong it is its camera designer call...
just look at this ...
gh2 vs olympus e-5
http://fourthirds-user.com/2011/03/so_which_camera_is_the_resolution_king.php
12 MP beats 16 MP...
e-5 has wery light olpf..
this is just example not fun boy talk...
but when u remove olpf u get more high frequencies details and they are very hard to compressed ...
codec is probably more stressed ....
 
Jordi Luque said:
Happy to play with James Miller's 5Dmk3 Raw footage for post-process test upscaling to 444 12bit colorspace. Please take a look:
http://vimeo.com/39755658
Click to expand...

Sorry mate but the saturation and sharpness makes it look like video, the original looks filmic. There's a halo effect starting to appear around peoples heads. Have a look at the guy with the sunglasses as he passes in front of the sign. It's not that the original isn't sharp as such, it's just the shallow DOF that makes it look soft.
 
jcs said:
nothing- see the tear down pics- there's a connector.
bernard- connector is for dust vibration cleaning system.
Click to expand...

Any way to keep that extremely useful cleaning system? Maybe by changing the OLPF for a clean pice of glass?

John, I missed that post of you in my first read. This thread advances at lightspeed... :

It should be possible to remove the vibration system and attach to a new piece of glass. If it's glued, etc., then perhaps a new component can be made and the process copied.
Click to expand...


 
Last edited:
jcs said:
I'd like to see what happens by just disconnecting the 2nd OLPF connector (and leaving the OLPF in place, in the event the firmware behaves differently when the connector is removed).
Click to expand...

I agree, I'd be interested to see what happens as well! Something to think about it, what if Canon is turning on the dust reduction system during video, they could have tuned the frequency to reduce image sharpness at a larger frequency than what is needed for still images. Disconnecting it could see if the results in video are the same or difference, which would isolate the glass versus the connector. I like it.
 
I'd like to see a comparison with 5D3 with no OLPF compared with Nikon D800. I like Nikon's skin tones and res, but I really wanna see what the aliasing is like...
 
eleison said:
Great :-( I just ordered the d800... Now this comes up. Back to the drawing boarding. I wish canon would play it straight instead of milking its products.

Who knows. I might just keep the d800 when it arrives (the d800 wouldn't arrive until early june because of the high demand which is probably a good thing). So far, IMHO, nikon has played it "more straight" than canon. Hopefully, the clean hdmi thing works out. I'm really sick of canon and would love to switch systems. I only have a 7d and several great lens so switching wouldn't be such a big deal... but man, I was all set on switching and giving canon "the bird".
Click to expand...
Well, if you can't cancel, (if you'd even need or want to), hopefully, the 14+stops of DR in iso100-200 is available in the video mode for some solace.
 
Why don't you guys go out and shoot something instead of pixel peeping yourself in circles? This has turned into the most ridiculous thread ever, people obsessing over the insides of someone else's camera and hypothesizing BS about every connector.

An OLPF is just a piece of glass. It blurs, nothing more, nothing less. People have been removing them for years for increased sharpness on many different cameras, nothing new here.

There is no deliberate corporate crippling happening - just possibly poor decision making in the quest to be everything to everyone. Not surprising that still photography customers' needs win out on a stills camera.

Get out and shoot.
 
Last edited:
nothing said:
Why don't you guys go out and shoot something instead of pixel peeping yourself in circles? This has turned into the most ridiculous thread ever, people obsessing over the insides of someone else's camera and hypothesizing BS about every connector.

Get out and shoot.
Click to expand...

Dear nothing. Your username is very apt for the contribution value of your post.

Please don't assume the people here are not going out and shooting.


Just saying :violin:
 
@nothing
this kind of forums are always part OCD country part wild guesses ....
learn to ignore that and u will find a lot of useful information ...
PS:thanks for contribution you are spot on ...
 
Last edited:
if canon 5d mark does 2X2 binning (not 3x3 requires different sensor design )
image coming from sensor are not 16MP in 16:9 they are 4MP ...
what canon digic 5 does is resizing 4mp to HD by some quick dirty bilinear method ...
that is reason why we dont see perfect 1000 lp...
but if aa filter is removed than there is more resolution squeezed ...
alexa 4.5 MP sensor cant resolved 1000lp its more 850lp ...
even f35 with 12 mp resolves 750 because binning...
so canon binning should gives as about 800 line ...
remove olpf and we can get close to 1000....
 
You must log in or register to reply here.
Back
Top